XLSX: header/footer font and size codes are ignored, so every printed page carries the fallback serif

[developer0hye]

Summary

headerFooter/oddHeader and oddFooter strings carry inline formatting codes. &"<font>,<style>"
selects the face and style for the runs that follow it, and there is a separate &<n> size code.
office2pdf renders header and footer text in the fallback serif at the document default size
instead, so the family and the size are both wrong on every printed page.

Measured

headerFooterTest.xlsx sets &L&"Calibri,Regular"&K000000top left and five more slots
(left/centre/right of both header and footer).

	GT (Excel)	office2pdf
family	Calibri	LibertinusSerif-Regular
size	12pt	11pt
slots affected	6 of 6	6 of 6

The serif-vs-sans difference is plainly visible at 150 DPI, and it appears on every page of
any workbook that sets a header or footer.

Worth noting alongside it: the same GT PDF renders its cell text in a substituted Malgun
Gothic while the header/footer text is real Calibri 12 — so the 12pt figure above is Excel's
own header font size, not an artifact of the substitution.

Reproduction

cargo run -p office2pdf-cli -- tests/fixtures/xlsx/poi/headerFooterTest.xlsx -o out.pdf
mutool draw -F stext -o - out.pdf 1 | grep -i 'font name'   # LibertinusSerif 11, Excel uses Calibri 12

Expected

&"<font>,<style>" selects the named family and style for the following runs, &<n> sets
their size, and a slot with no explicit codes uses Excel's default header/footer font
(Calibri 12 in this file's theme) rather than the document body fallback.

Actual

Both codes are ignored; every slot renders LibertinusSerif-Regular at 11pt.

Related: #233, #578

Found in the 2026-07-29 full-corpus visual audit at c1adfbd (native Office GT, 150 DPI, mutool-trace geometry).

Visual comparison

Compare

GT left, office2pdf right, 150 DPI, same page.

[developer0hye]

Fixed in #799.

parse_hf_format_string was parsing both codes and discarding them — the branches were "skip to closing quote" and "skip digits" — so every header and footer printed in the fallback serif. Each section is now a sequence of style-carrying segments, so a code partway through a section applies from that point rather than to the whole slot.

Measured on tests/fixtures/xlsx/headerFooterTest.xlsx, whose six slots each set &"Calibri,Regular". Embedded fonts go from LibertinusSerif-Regular + Carlito-Regular to Carlito-Regular alone — cell text was already Carlito, so the serif was the header/footer and nothing else.

One part of this report is not fixed, and I want to be explicit rather than let the close imply otherwise: the 11pt-vs-12pt size difference. That fixture sets no &<n> code at all, so its size does not come from a discarded code — it is Excel's default header size, which is a separate question. The &<n> path is implemented and unit-tested; if the size difference still shows against a ground truth, please reopen or file it separately and I will take it on its own evidence.